def summarize()

in src/text_clustering.py [0:0]

• 21 lines of code
• 4 McCabe index (conditional complexity)

    def summarize(self, texts, labels):
        unique_labels = len(set(labels)) - 1  # exclude the "-1" label
        client = InferenceClient(self.summary_model, token=self.summary_model_token)
        cluster_summaries = {-1: "None"}

        for label in range(unique_labels):
            ids = np.random.choice(self.label2docs[label], self.summary_n_examples)
            examples = "\n\n".join(
                [
                    f"Example {i+1}:\n{texts[_id][:self.summary_chunk_size]}"
                    for i, _id in enumerate(ids)
                ]
            )

            request = self.summary_template.format(
                examples=examples, instruction=self.summary_instruction
            )
            response = client.text_generation(request)
            if label == 0:
                print(f"Request:\n{request}")
            cluster_summaries[label] = self._postprocess_response(response)
        print(f"Number of clusters is {len(cluster_summaries)}")
        return cluster_summaries